Data hygiene - is your data clean?

27.12.2024

Rapid advancements in sports technology promise performance gains but risk data overload. Prioritising data hygiene ensures accuracy and reliability for better decisions!

1. Sports tech boom!

In an overwhelming world of ever-increasing exposure to sports technology, the phrase “sinking in the data” becomes ever so relevant. Sports technology has boomed in the 2000’s, with everybody looking for that extra 1% on their competitors. Companies have found their place in the industry with bold claims for their ergogenic aids, injury prediction tools and recovery enhancing modalities. Some have done it well, boasting a rich and organic collection of scientific underpinning prior to production, or even prior to financial realisation. Yet some big tech giants come across as just another sales team with a shiny gadget built for mass production. Often built with minimal scientific underpinning, limited practical experience of its true application and suitability for their desired fields – I won’t mention any names.

Sports technology should inform, not impede our practice. Information overload equals filter failure, and a wealth of information often leads us to a poverty of attention. Sports tech to a coach is what a protein shake is to an athlete, a supplement to enhance a skillset, not an over-reliant and under-effective drug. It’s important to be incredibly critical of sports technology, especially when mass produced by corporate giants who often have never stepped foot into your field, or lab rats who think they have the answers but have no environmental validity outside of their four white walls and medical gowns. Just because something looks good, it doesn’t necessarily mean it performs well, or provides you with reliable, valid, and the representative data that you are looking for. Look out for swiss army knives, jack of all trades, master of none. Convenience should not impede quality! A fold-away product? Fantastic, but does it hold its own when any sort of reasonable force is applied to it? I know which I’d rather. It’s like that toy that does so many things, but none of them very well - it ends up collecting dust. Our sports tech tools should be research evidenced and operationally robust, so that we have clarity and confidence with our data capture. And don’t get led down the garden path assessing and training every single thing under the sun, just because you can. Rationalise your test selection, operationalise reliable procedures, focus big rocks, and always have perspective on what really matters. After all, if we are looking for the marginal gains, surely our room for error is going to be pretty small?

2. Data hygiene

2.a. Drawing back the data curtains

Below is a force-time curve from a maximal voluntary isometric contraction (MVIC) collected by our friends at King Performance Ideology. 

Key metrics that practitioners obtain from these time curves are presented below:

  • Peak force, average force and limb asymmetry, often in Newton’s (N).
  • Peak rate of force development, often in Newton’s per second (N/s).
  • Rate of force development at specific time points, often 0-50, 50-100, 100-150, 150-200 and 200-250 ms time windows.

Often, these numbers are presented to athletes, fellow coaches, scouts, management, and executive staff members at elite sports organisations. This data is concise and easy to interpret when presented alongside normative data, or previous bests, or simply some colour coding to outline whether something is good (green), average (yellow), or bad (red). Important decisions are often made following this kind of information, sometimes surrounding selection for a competitive fixture, readiness to return from a significant long-term injury, and in some cases for talent identification purposes undertaken by recruitment departments at elite sports organisations. That’s a lot of pressure riding on the shoulders of our data, right?!

Peeling back the layers behind the single resultant score are several potentially confounding factors that may either increase or decrease the resultant quality and accuracy of the data. We define this as “data hygiene”. At Metrics, we use the following equation to determine data hygiene, and to provide clarity on what story the observed score truly tells:

Data hygiene = true score - observed score

A true score is humanly impossible to obtain due to the inevitable error that exists in us as human beings and within the equipment that we build and adopt. Therefore, the observed score can be viewed as the true score minus any induced error that exists within the data collection process. The smaller the data hygiene score, the cleaner and more accurate the final data.

2.b. Threats to data hygiene

Minimising the gap between a true score and an observed score requires meticulous planning and care. Let’s take a look at all of the potential threats to data hygiene during the data manufacturing process:

Final data output > expected overall error > systematic error of the equipment (electronics within hardware, apparatus failure) > athlete motivation > athlete education > administrator rigour > administrator education level > athlete preparation (warm up) > test preparation (angles) > athlete training status > device calibration status > device manufacturing quality > movement specific relevance / underpinning.

Each of these threats can be bucketed into specific categories that allow us to better understand the quality of our devices, and the areas in which error may be derived from.

  • Human error. This involves factors surrounding administrator skill level, training status, setup accuracy and rigour, and athlete motivation levels, preparation and training status.
  • Systematic bias. This outlines factors such as athlete training status, education level, fatigue status, and timing of data collection.
  • Apparatus quality. These largely govern factors relating to the system, such as calibration status, manufacturing quality, biomechanical relevance and equipment ergonomics.

2.c. Data hygiene mistakes

Below are a few common mistakes that can be made when utilising sports technology tools, specifically those of the popular isometric strength kind.

Human error - joint angles.

Manual goniometers, and eyeballing joint angles with the naked eye can be incredibly inaccurate and provide inconsistent and unreliable standards for participant preparation. In Tom King’s PhD, joint angle alterations of just 15° elicits a change in force production and muscle excitation.

Here’s how joint angle impacts force production:

  • Muscle length-tension relationship. Each muscle has a specific “optimal muscle length” for maximum force production, which is at the point where there is an ideal overlap between actin and myosin filaments within the sarcomeres. Either side of this optimum length causes either excessive or insufficient overlap, and a reduction in the number of cross-bridges.
  • Moment arm. This is the perpendicular distance from the axis of rotation (joint center) to the line of action of the muscle force. A longer moment arm produces more torque.
  • Mechanical properties. The stiffness of a muscle-tendon unit affects how force is  transmitted. At certain joint angles, tendons can also store elastic energy when stretched and release it to assist in force production.

Systematic bias - familiarisation.

Time-constrainted environments in competitive sport minimise the opportunities for diagnostic screening with athletes. Therefore, dedicated education and familiarisation sessions for athletes to become accustomed to diagnostic equipment prior to maximal testing are largely nonexistent.

The problem proliferates when data collected during testing of a new athlete at a time like pre-season is used in later circumstances (e.g. as a pre-injury marker of strength, or as a means to understand changes after a period of training). At this point, without substantial familiarisation, it becomes hard to determine whether test 1 (e.g. pre-season) is greater or less than test 2 (e.g. post-injury or end of training programme) because of a true change, or simply due to a learning effect.

Apparatus quality - equipment ergonomics.

We’ve all tested an athlete and heard the worst “it hurts so I don’t want to push harder” come out of their mouth. Unfortunately, this isn’t just a problem of the athlete needing to suck it up and push through - largely this is due to the lack of ergonomic appreciation that equipment should hold when claiming to test a certain physical quality. Do we really think a small slightly padded disk on the top of your knee (hip flexor) or on the inside of your knee (adductor) is going to be the best tool to ask our athletes to push maximally? Probably not.

Our in-house analysis has discovered that utilising a padded strap to “pull” vs. a small disk to “push” can evoke an improvement in results of up to 24%!!! That’s a huge amount of force missed simply due to discomfort. *Observed score vs. true score reminder.

Data Table
Test type Force Difference
Iso-Push Disc 410 N 24%
Iso-Pull Straps 510 N -

2.d. Best practice for data hygiene

At Metrics, we believe in setting new standards for data hygiene in sports technology. Without it, we will never improve our ability as practitioners to better inform training prescription, improve physical performance, and crack the injury crisis that so many sports are experiencing in modern day competitive sport.

In order to achieve greater success in the field, we have built a simple 5 step process for practitioners to adopt within their sporting environments. Keep an eye out for our publication of this process in the coming weeks, and click here to join an email waiting list to be first to get your hands on it!

Register your interest

Discover how Metrics can help you optimise your potential. Contact Us today to learn more.

Thank You for Submission

You can expect a response within 48 hours. Stay tuned for some news!

Something went wrong! Please double-check your submission and try again.